Deep Reinforcement Learning‐Driven Cooperative Routing for Energy Efficiency in Wireless Multimedia Sensor Networks

ABSTRACT The rapidly evolving landscape of cooperative routing protocols in network systems necessitates innovative approaches to address their inherent challenges, such as energy depletion, scalability limitations, QoS maintenance difficulties, and dynamic topology changes. This research introduces a deep reinforcement learning (DRL) framework designed to enhance the efficiency and effectiveness of cooperative routing protocols. A relay‐based routing mechanism is developed using the proposed enhanced deep reinforcement learning (EnDRL) technique. In this approach, the DRL framework leverages the adaptive kookaburra optimization (AdKo) algorithm to select the optimal action, significantly improving routing efficiency. The AdKo algorithm is formulated to select the optimal action by incorporating adaptive concepts into the traditional kookaburra optimization algorithm. This adaptation involves dynamically adjusting weights to improve the convergence rate and mitigate the risk of local optimal trapping, thereby enhancing the overall performance of the optimization process.